Efficacy of Olanzapine in Combination with Valproate Or Lithium in the Treatment of Mania in Patients Partially Nonresponsive to Valproate Or Lithium Monotherapy

ORIGINAL ARTICLE Efficacy of Olanzapine in Combination With Valproate or Lithium in the Treatment of Mania in Patients Partially Nonresponsive to Valproate or Lithium Monotherapy

Mauricio Tohen, MD, DrPH; K. N. Roy Chengappa, MD; Trisha Suppes, MD, PhD; Carlos A. Zarate, Jr, MD; Joseph R. Calabrese, MD; Charles L. Bowden, MD; Gary S. Sachs, MD; David J. Kupfer, MD; Robert W. Baker, MD; Richard C. Risser, MSc; Elisabeth L. Keeter, RN, MSN; Peter D. Feldman, PhD; Gary D. Tollefson, MD, PhD; Alan Breier, MD

Background: A 6-week double-blind, randomized, pla- proved 21-item Hamilton Depression Rating Scale (HAMD- cebo-controlled trial was conducted to determine the ef- 21) total scores significantly more than monotherapy (4.98 ficacy of combined therapy with olanzapine and either vs 0.89 points; PϽ.001). In patients with mixed-episodes valproate or lithium compared with valproate or lithium with moderate to severe depressive symptoms (DSM-IV alone in treating acute manic or mixed bipolar episodes. mixed episode; HAMD-21 score of Ն20 at baseline), olanzapine cotherapy improved HAMD-21 scores by 10.31 Methods: The primary objective was to evaluate the ef- points compared with 1.57 for monotherapy (PϽ.001). Ex- ficacy of olanzapine (5-20 mg/d) vs placebo when added trapyramidal symptoms (Simpson-Angus Scale, Barnes Aka- to ongoing mood-stabilizer therapy as measured by re- thisia Scale, Abnormal Involuntary Movement Scale) were ductions in Young Mania Rating Scale (YMRS) scores. not significantly changed from baseline to end point in ei- Patients with bipolar disorder (n=344), manic or mixed ther treatment group. Treatment-emergent symptoms that episode, who were inadequately responsive to more than were significantly higher for the olanzapine cotherapy group 2 weeks of lithium or valproate therapy, were random- included somnolence, dry mouth, weight gain, increased ized to receive cotherapy (olanzapine + mood- appetite, tremor, and slurred speech. stabilizer) or monotherapy (placebo + mood-stabilizer). Conclusion: Compared with the use of valproate or Results: Olanzapine cotherapy improved patients’ YMRS lithium alone, the addition of olanzapine provided su- total scores significantly more than monotherapy (−13.11 perior efficacy in the treatment of manic and mixed bi- vs −9.10; P=.003). Clinical response rates (Ն50% improve- polar episodes. ment on YMRS) were significantly higher with cotherapy (67.7% vs 44.7%; PϽ.001). Olanzapine cotherapy im- Arch Gen Psychiatry. 2002;59:62-69

HE EXPERT Consensus chotic therapy seems superior to antipsy- Guidelines Series, pub- chotic agents alone.5 In support of this, lished in the year 2000, Mu¨ ller-Oerlinghausen et al6 compared the recommends lithium and efficacy of combined therapy with con- valproate as first-line treat- ventional antipsychotics and valproate vs mentsT for bipolar mania.1 However, up to valproate monotherapy in patients with bi- 40% of patients respond poorly to mono- polar or schizoaffective disorder and found therapy with either treatment.2 When mono- combination therapy to be superior to therapy fails, the guidelines recommend monotherapy. combination therapies. A number of au- Olanzapine, an atypical antipsy- thors have recently reviewed the use of such chotic, has been shown in 2 placebo- cotherapies for bipolar mania. Freeman and controlled studies to have acute antimanic A list of the Principal Stoll3 concluded that the combination of effects.7,8 Moreover, a previous report has Investigators appears in lithium and valproate is better tolerated and suggested that olanzapine is effective when the box on page 69. Author more efficacious in maintenance therapy used in combination with other psycho- affiliations appear in the 9 acknowledgment section. than other combination treatments. tropic agents. The present study was con- Drs Tohen, Feldman, Tollefson, Typical neuroleptics have been sug- ducted to investigate the efficacy and safety and Breier and Mr Risser and gested to be superior in efficacy to lithium of combined therapy with olanzapine and Ms Keeter are stockholders monotherapy.4 Conversely, the addition of either valproate or lithium compared with in Eli Lilly & Co. a mood stabilizer to conventional antipsy- valproate or lithium monotherapy.

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 59, JAN 2002 WWW.ARCHGENPSYCHIATRY.COM 62

SUBJECTS Patient assessments were conducted by mental health care professionals, including psychiatrists, psychologists, nurses, All patients were diagnosed as having bipolar disorder, manic and other mental health caregivers with a clinical degree or mixed episode, with or without psychotic features, us- or certification. Raters were trained in the use of the SCID ing the Structured Clinical Interview for the DSM-IV10 and symptom-rating scales before study initiation. To en- (SCID).11 Patients had to have at least 2 previous de- sure high interrater reliability, investigators were re- pressed, manic, or mixed episodes as well as a Young Ma- quired to achieve a reliability coefficient of 0.75 or greater. nia Rating Scale12 (YMRS) total score of 16 or greater at visit The primary measure of efficacy to assess severity of manic 1 and visit 2 (2-7 days later). Patients were required to have symptoms was the mean change from baseline to end point had a documented trial of treatment, with a therapeutic in the YMRS total score. Secondary measures included the blood level of lithium (0.6-1.2 mmol/L) or valproate (50- 21-item Hamilton Depression Rating Scale13 (HAMD-21); 125 µg/mL), for at least 2 weeks immediately prior to visit the Positive and Negative Syndrome Scale14; and the Clini- 1. Patients were included only if they showed inadequate cal Global Impressions Severity of Bipolar Disorder scale14 response to monotherapy (YMRS total score Ն16). Prior (CGI-BP) total scores, and mania and depression subscale to participation, all patients signed an informed consent scores. Clinical responses on the YMRS and HAMD-21 were document approved by their study site’s institutional re- defined a priori as an improvement of 50% or greater. Clini- view board. cal remission (euthymia) was defined a priori as achievement of a YMRS total score of less than or equal to 12. A STUDY DESIGN subsample of patients with moderate to severe depressive symptoms was defined by a current mixed episode and a Participants in the study initially entered a 2- to 7-day screen- HAMD-21 total score of 20 or greater at baseline. Second- ing and washout period (study period 1) during which all ary assessments, also defined a priori, included analyses of concomitant medications other than lithium or valproate treatment differences following stratification by the cur- were discontinued. Patients already receiving valproate or rent course of illness, the presence or absence of psy- lithium continued to do so throughout the study. Patients chotic features, and the use of lithium or valproate. receiving other forms of treatment started receiving either Scales for the assessment of neurologic adverse events lithium or valproate at investigator discretion for the 2 weeks included the Simpson-Angus Scale,15 the Barnes Akathisia immediately prior to visit 1. Plasma levels of the medica- Scale,16 and the Abnormal Involuntary Movement Scale.14 tions were documented to be within the therapeutic ranges. Assessment of vital signs, weight, and clinical laboratory Only patients scoring greater than or equal to 16 on the analytes (including prolactin, nonfasting glucose, and elec- YMRS were randomized to receive concurrent treatment trolyte levels and hematologic analysis) was performed at combined with either olanzapine or placebo (study pe- each visit. Serum concentrations of mood stabilizers were riod 2). collected at every visit. Study period 2 consisted of a 6-week acute, double- blind phase, during which levels of lithium or valproate were STATISTICAL ANALYSES maintained within the therapeutic range. Patients were as- sessed weekly. Patients were randomized 2:1 to receive ei- Data were analyzed on an intent-to-treat basis,17 included ther olanzapine (flexible dose range of 5, 10, 15, or 20 mg/d) all patients who met the entry criteria (including inad- added to valproate or lithium or placebo added to valpro- equate responsiveness to the minimum 2-week prior treat- ate or lithium. Olanzapine therapy was initiated at 10 mg/d. ment with lithium or valproate), and provided both a base- To maintain blinding, treatment took the form of two 5-mg line and at least 1 postbaseline data measurement. Total capsules (either olanzapine or placebo), titrated up in in- scores from rating scales were derived from the individual crements of 1 capsule or down by any number of decre- items; if any item was missing, the total score was treated ments at investigator discretion as indicated by each pa- as missing. All tests were 2-sided, with an ␣ level of .05. tient’s tolerance. Patients unable to tolerate the minimum Analysis of variance (ANOVA) models were used to evalu- dose were discontinued. Patients were permitted adjunc- ate continuous data, including terms for treatment, inves- tive use of benzodiazepine (Յ2 mg/d of lorazepam equiva- tigator, and treatment-investigator interaction. The linear lents) for no more than 14 days cumulatively. Anticholin- model for this analysis included terms for baseline, treat- ergic therapy (benztropine mesylate, Յ2 mg/d) was ment, investigator, treatment-investigator interaction, visit, permitted throughout the study for treatment of extrapy- and treatment-visit interaction. The Fisher exact test was ramidal symptoms but not for prophylaxis. Aside from study used for categorical analyses, including laboratory values, drugs, benzodiazepines, and anticholinergics, no other drugs vital signs, and treatment-emergent adverse events. Data were permitted during the study. are given as mean (SD) unless otherwise indicated.

RESULTS per site. Patients were recruited from both academic and nonacademic sites from existing clinical patient popu- PATIENT CHARACTERISTICS AND DISPOSITION lations seeking treatment at those sites. Of the 344 randomized patients, 322 came from outpatient centers. The A total of 501 patients entered the screening phase and other 20 (cotherapy, n=16; monotherapy, n=4) came 344 patients were randomized and enrolled (33 US cen- from inpatient settings. Patients were initially screened ters, 5 Canadian), with a mean enrollment of 9 patients on the basis of face-to-face interviews, medical record re-

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 59, JAN 2002 WWW.ARCHGENPSYCHIATRY.COM 63

Olanzapine Cotherapy Monotherapy Characteristic Mood Stabilizer (n = 229) (n = 115) P Value* Age, mean ± SD, y 40.7 ± 11.2 40.4 ± 10.8 .9 Lithium 40.8 ± 12.4 43.4 ± 11.0 .2 Valproate 40.7 ± 10.7 38.9 ± 10.5 .25 Male, No. (%) 101 (44.1) 64 (55.6) .05 Lithium 41 (54.0) 26 (63.4) .34 Valproate 60 (39.2) 38 (52.1) .08 White, No. (%) 196 (85.6) 97 (84.4) .75 Lithium 65 (85.5) 34 (82.9) .79 Valproate 131 (85.6) 63 (86.3) Ͼ.99 Current course, No. (%) .21 Manic 104 (45.4) 61 (53.0) Mixed 125 (54.6) 54 (47.0) Lithium .03 Manic 38 (50.0) 12 (29.3) Mixed 38 (50.0) 29 (70.7) Valproate Ͼ.99 Manic 87 (56.9) 42 (57.5) Mixed 66 (43.1) 31 (42.5) Without psychotic features, No. (%)† 154 (67.3) 76 (66.1) .90 Lithium 54 (71.1) 28 (68.3) .83 Valproate 100 (65.4) 48 (65.8) Ͼ.99 Cotherapeutic agent, No. (%) .63 Lithium 76 (33.2) 41 (36.0) Valproate 153 (66.8) 73 (64.0)

*Treatment difference, olanzapine cotherapy vs monotherapy; derived from analysis of variance for age and from the Fisher exact test otherwise. †Based on n = 114 for monotherapy.

views, and information obtained from family members manic episodes. Overall baseline mean YMRS total scores and referring clinicians. Reasons for lack of enrollment for the olanzapine cotherapy (n=220) and mono- included entry criteria not met (86 patients, including therapy (n=114) groups were 22.31 (5.39) and 22.67 24 failing to meet the YMRS total score criterion of Ն16); (5.15), respectively, and mean HAMD-21 scores were patient decision or loss to follow-up during the screen- 14.52 (8.46) and 13.54 (7.63), respectively (Table 3). ing phase (58); investigator decision (8); protocol vio- Mean modal dose of olanzapine in the cotherapy lation (4); and a single death that occurred before group (n=224) was 10.4 (4.9) mg/d). Mean plasma lev- completion of screening or exposure to the study drug. els of lithium among the cotherapy (n=74) and mono- Ultimately, 229 patients were randomized to receive therapy (n=41) patients were 0.76 (0.16) and 0.82 (0.19) olanzapine cotherapy and 115 to receive monotherapy (F1,86=4.26; P=.04) mEq/L, respectively, while mean (Table 1). One patient in the monotherapy group re- plasma levels of valproate for cotherapy (n=145) and ceived both valproate and lithium and accordingly was monotherapy (n=73) were 63.6 (18.4) µg/mL and 74.7 Ͻ excluded from the subgroup analyses. The median du- (18.6) µg/mL, respectively (F1,188=18.38; P .001). Ben- ration of mood-stabilizer therapy prior to randomiza- zodiazepine use was not statistically different between tion was 67 days; 203 patients had a duration of therapy patients in the cotherapy (66/229 [28.8%]) and mono- longer than 6 weeks. One patient in the monotherapy therapy (39/115 [33.9%]) groups (P=.38). group and 9 in the cotherapy group had no postbaseline measures and were excluded from all efficacy analyses. PRIMARY OUTCOMES The percentage of patients completing the study was roughly equal in the 2 treatment groups (cotherapy, Both groups of patients improved during the course of 69.9%; monotherapy, 71.3%). Significantly more pa- treatment as indicated by the primary measure of effi- tients in the monotherapy group discontinued treat- cacy, the YMRS total score (Table 3). However, the olan- ment due to lack of efficacy (12.2% vs 3.1%; P=.002), zapine cotherapy group (n=220) showed a mean de- whereas significantly more patients in the cotherapy group crease in YMRS total score of 13.1 (8.53) points, withdrew due to adverse events (10.9% vs 1.7%; P=.002) corresponding to a 58.8% improvement from baseline (Table 2). compared with a decrease of 9.10 (9.36) points F1,276=9.08; Patient demographics and illness characteristics were P=.003) for the monotherapy group (n=114), which cor- not significantly different between the cotherapy and responded to an improvement of 40.1%. monotherapy treatment groups overall (Table 1). In the Itemwise analysis of the YMRS revealed that, com- overall study group (n=344), the mean age was 40.6 pared with monotherapy, olanzapine cotherapy brought (11.1) years. One hundred sixty-five patients (48.0%) had about significantly greater improvement at end point on mixed episodes at enrollment; the remainder had pure the items of Irritability (cotherapy, −1.82 [2.09], n=220;

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 59, JAN 2002 WWW.ARCHGENPSYCHIATRY.COM 64

Olanzapine Cotherapy Monotherapy Characteristic Mood Stabilizer (n = 229) (n = 115) P Value† Completed study, No. Full sample 160 (69.9) 82 (71.3) .80 Lithium 58 (76.3) 32 (78.1) Ͼ.99 Valproate 102 (66.7) 50 (68.5) .88 Reasons for discontinuation Adverse event Full sample 25 (10.9) 2 (1.7) .002 Lithium 5 (6.6) 0 .16 Valproate 20 (13.1) 2 (2.74) .02 Lack of efficacy Full sample 7 (3.1) 14 (12.2) .002 Lithium 2 (2.6) 4 (9.8) .18 Valproate 5 (3.3) 9 (12.3) .02 Lost to follow-up Full sample 5 (2.2) 3 (2.6) Ͼ.99 Lithium 2 (2.6) 2 (4.9) .61 Valproate 3 (2.0) 1 (1.4) Ͼ.99 Patient decision Full sample 13 (5.7) 2 (1.7) .16 Lithium 3 (4.0) 1 (2.4) Ͼ.99 Valproate 10 (6.5) 1 (1.4) .12 Criteria not met/compliance Full sample 12 (5.2) 5 (4.3) .8 Lithium 3 (4.0) 1 (2.4) Ͼ.99 Valproate 9 (5.9) 4 (5.5) Ͼ.99 Sponsor decision Full sample 1 (0.4) 1 (0.9) Ͼ.99 Lithium 0 1 (2.4) .35 Valproate 1 (0.7) 0 Ͼ.99 Physician decision Full sample 5 (2.2) 6 (5.2) .19 Lithium 2 (2.6) 0 .54 Valproate 3 (2.0) 6 (8.2) .06 Satisfactory response Full sample 1 (0.4) 0 Ͼ.99 Lithium 1 (1.3) 0 Ͼ.99 Valproate 0 0

*Data are given as number (percentage) unless otherwise indicated. †Treatment difference, olanzapine cotherapy vs monotherapy; derived from the Fisher exact test.

monotherapy, −1.02 [2.37], n=114; F1,276=5.69; P=.02); with a median remission time of 14 days for cotherapy Speech (cotherapy, −2.45 [2.03], n=220; monotherapy, vs 22 days for monotherapy. −1.63 [2.53], n=114; F1,276=5.24; P=.02); Language/ Compared with the patients in the monotherapy group, Thought Disorder (cotherapy, −0.94 [0.91], n=220, patients in the olanzapine cotherapy group showed sig- monotherapy, −0.72 [1.00], n=114; F1,276=5.34; P=.02); nificantly greater improvement on the HAMD-21 at each and Disruptive/Aggressive Behavior (cotherapy, −1.18 time point throughout the study. By week 6, the co- [1.64], n=220; monotherapy, −0.46 [1.77], n=114; therapy group (n=220) experienced a mean last observa- F1,276=10.16; P=.002). tion carried forward decrease in HAMD-21 scores of 4.98 Ͻ Clinical response was defined a priori in the protocol (7.61) points, significantly greater (F1,276=18.05; P .001) as improvement of 50% or greater from baseline to end point than the decrease of 0.89 (6.90) points in the mono- in the YMRS total score. On this basis, 149 (67.7%) of the therapy group (n=114). An exploratory itemwise analy- 220 patients in the olanzapine cotherapy group re- sis showed significantly greater improvement in the di- sponded to treatment compared with 51 (44.7%) of the 114 mensions of depressed mood, feelings of guilt, suicidality, patients in the monotherapy group (PϽ.001). In addi- early insomnia, anxiety-psychic, and paranoid symptoms. tion, time to response was significantly shorter for co- Analysis of end point HAMD-21 scores conducted in therapy (P=.002, log rank test), with a median response the subset of patients experiencing a mixed episode with time of 18 days for cotherapy vs 28 days for monotherapy. moderate to severe depressive symptoms at baseline (HAMD-21 total score Ն20 at baseline) showed a de- SECONDARY OUTCOMES crease of 10.31 (8.19) points for olanzapine cotherapy (n=51) compared with 1.57 (7.73) points (F1,70=17.50; Clinical remission was defined a priori in the protocol PϽ.001) for monotherapy (n=21). Within this subset, as achievement of a YMRS total score of less than or equal 43.1% of patients in the cotherapy group showed Ն50% to 12. On this basis, 173 (78.6%) of the 220 patients in improvement of depressive symptoms compared with 9.5% the olanzapine cotherapy group demonstrated evidence in the monotherapy group (P=.006). of remission. In the monotherapy group, 75 (65.8%) of Other secondary measures of efficacy included the the 114 evaluated patients demonstrated evidence of re- Positive and Negative Syndrome Scale (total; Positive, mission. This difference in remission rates was also sig- Negative, and Cognitive clusters; and Hostility sub- nificant (P=.01). Time to remission was significantly scores) and the CGI-BP (overall, Severity of Mania, and shorter in the cotherapy group (log rank test, P=.002), Severity of Depression). Olanzapine cotherapy brought

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 59, JAN 2002 WWW.ARCHGENPSYCHIATRY.COM 65

Monotherapy Olanzapine Cotherapy (n = 220) (n = 114) Comparison

Measurement Mood Baseline: Change: Baseline: Change: F Effect P Scale Stabilizer mean (SD) mean (SD) mean (SD) mean (SD) Statistic† Size Value‡ YMRS Total Full sample 22.31 (5.39) −13.11 (8.53) 22.67 (5.15) −9.10 (9.36) 9.08 0.47 .003 Lithium 22.34 (5.26) −13.62 (8.36) 22.22 (4.65) −10.39 (8.69) 3.10 0.42 .08 Valproate 22.29 (5.48) −12.85 (8.64) 22.76 (5.31) −8.39 (9.76) 4.51 0.50 .04 HAMD-21 Total Full sample 14.52 (8.46) −4.98 (7.61) 13.54 (7.63) −0.89 (6.90) 18.05 0.58 Ͻ.001 Lithium 14.26 (8.33) −4.15 (8.18) 10.90 (6.54) −1.32 (5.19) 3.58 0.40 .06 Valproate 14.65 (8.55) −5.40 (7.30) 15.04 (7.89) −0.67 (7.77) 14.77 0.65 Ͻ.001 CGI-BP Overall Full sample 4.10 (0.74) −1.20 (1.16) 4.18 (0.72) −0.89 (1.31) 4.48 0.27 .04 Lithium 4.12 (0.78) −1.23 (1.24) 4.00 (0.71) −0.98 (1.44) 0.24 0.20 .62 Valproate 4.10 (0.73) −1.19 (1.12) 4.28 (0.72) −0.82 (1.24) 2.08 0.33 .15 CGI-BP Mania Full sample 4.06 (0.79) −1.48 (1.25) 4.13 (0.70) −1.16 (1.39) 2.94 0.26 .09 Lithium 4.12 (0.72) −1.61 (1.23) 4.02 (0.69) −1.10 (1.55) 1.70 0.40 .2 Valproate 4.03 (0.82) −1.42 (1.26) 4.18 (0.70) −1.18 (1.31) 0.10 0.20 .75 CGI-BP Depression Full sample 2.76 (1.40) −0.50 (1.33) 2.62 (1.37) 0.12 (1.45) 13.84 0.48 Ͻ.001 Lithium 2.59 (1.32) −0.35 (1.46) 2.07 (1.06) 0.10 (1.16) 0.46 0.36 .5 Valproate 2.84 (1.43) −0.58 (1.26) 2.93 (1.44) 0.17 (1.58) 15.79 0.57 Ͻ.001 PANSS Total Full sample 62.10 (17.28) −12.90 (15.72) 61.75 (15.51) −6.96 (16.39) 8.78 0.42 .003 Lithium 61.43 (15.85) −14.03 (15.15) 58.63 (13.27) −9.02 (12.59) 0.93 0.40 .34 Valproate 62.45 (18.00) −12.34 (16.02) 63.31 (16.50) −5.86 (18.28) 7.80 0.42 .006 PANSS Cognition Full sample 14.36 (4.32) −3.08 (4.12) 14.50 (3.90) −2.29 (4.23) 2.59 0.21 .12 Lithium 14.12 (4.04) −3.39 (4.10) 14.41 (3.76) −3.37 (4.07) 0.01 0.01 .92 Valproate 14.49 (4.46) −2.92 (4.13) 14.50 (4.00) −1.72 (4.24) 4.98 0.31 .03 PANSS Hostility Full sample 9.54 (3.36) −2.99 (3.62) 9.58 (3.11) −1.69 (3.66) 4.95 0.39 .03 Lithium 9.57 (2.95) −3.49 (3.40) 9.49 (2.93) −2.05 (3.07) 3.66 0.45 .06 Valproate 9.53 (3.56) −2.73 (3.71) 9.57 (3.20) −1.49 (3.98) 1.80 0.39 .18

*YMRS indicates Young Mania Rating Scale; HAMD-21, Hamilton Depression Rating, 21-Item; CGI-BP, Clinical Global Impressions–Severity of Bipolar Disorder; and PANSS, Positive and Negative Syndrome Scale. †All tests based on 1 df. ‡Treatment difference, olanzapine cotherapy vs monotherapy; derived from analysis of variance.

about significantly greater improvement than monotherapy: −12.92 [8.37], n=121; monotherapy: −7.46 Ͻ therapy on patients’ last observation carried forward, Posi- [10.15], n=54; F1,146=17.31; P .001). However, among tive and Negative Syndrome Scale total, and Hostility item patients presenting with pure mania, the treatment dif- scores, as well as on the CGI-BP overall and Severity of ference did not achieve statistical significance (co- Depression scores (Table 3). therapy: −13.34 [8.77], n=99; monotherapy: −10.57 [8.40], n=60; F1,129=2.95; P=.09). The superiority of olan- SUBGROUP ANALYSES zapine cotherapy over monotherapy seen in patients with mixed episodes was found only in patients receiving val- Subgroup analyses, defined a priori, were conducted on proate (cotherapy: −13.18, [8.49], n=84; monotherapy: baseline to end point YMRS total scores. No significant −7.48 [10.74], n=42; F1,124=10.53; P=.002), whereas the interactions were seen between previous exposure to psy- treatment difference seen with lithium did not achieve chotropics (antidepressants, antipsychotics) and therapy statistical significance (cotherapy: −12.32 [8.15], n=37; (cotherapy, monotherapy). However, among all pa- monotherapy: −7.42 [8.14], n=12; F1,47=3.28; P=.08), tients without psychotic features, olanzapine cotherapy again despite the lack of association between course of was significantly more efficacious than monotherapy (co- illness and the differential effect of therapy (ANOVA test therapy: −13.25 [7.76], n=150; monotherapy: −8.32 of interaction, F1,274=0.14; P=.71). Ͻ [8.68], n=76; F1,196=16.97; P .001). Among patients Finally, among patients receiving valproate, olanza- without psychotic features, olanzapine cotherapy was pine cotherapy brought about significantly greater im- more effective than monotherapy regardless of whether provement in YMRS total scores compared with patients patients received lithium or valproate. However, among receiving valproate monotherapy (cotherapy: −12.85 [8.64], patients with psychotic features, responses to treatment n=146; monotherapy: −8.39 [9.76], n=72; F1,188=13.44; were not different between the cotherapy and mono- PϽ.001). Among patients receiving lithium, the greater therapy groups regardless of whether patients received improvement seen with olanzapine cotherapy relative to lithium or valproate—this despite the lack of associa- monotherapy did not achieve statistical significance (co- tion between the presence of psychotic features and the therapy: −13.62 [8.36], n=74; monotherapy: −10.39 [8.69], differential effect of therapy (ANOVA test of interac- n=41; F1,86=3.74; P=.06). The type of mood stabilizer was tion: F1,274=0.60; P=.44). not associated significantly with a differential effect of co- Among patients with a current mixed episode, olan- therapy compared with monotherapy (ANOVA test of in- zapine cotherapy was superior to monotherapy (co- teraction, F1,273=0.74; P=.39).

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 59, JAN 2002 WWW.ARCHGENPSYCHIATRY.COM 66

Olanzapine Cotherapy Monotherapy (n = 229) (n = 115) Event (COSTART Term) Mood Stabilizer (%) (%) P Value† Somnolence Full sample 51.5 27.0 Ͻ.001 Lithium 51.3 12.9 .03 Valproate 51.6 26.0 Ͻ.001 Dry mouth Full sample 31.9 7.8 Ͻ.001 Lithium 25.0 7.3 .02 Valproate 35.3 8.2 Ͻ.001 Weight gain Full sample 26.2 7.0 Ͻ.001 Lithium 21.1 4.9 .03 Valproate 28.8 8.2 Ͻ.001 Increased appetite Full sample 23.6 7.8 Ͻ.001 Lithium 15.8 4.9 .13 Valproate 27.5 9.6 .002 Tremor Full sample 23.1 13.0 .03 Lithium 25.0 9.8 .05 Valproate 22.2 15.1 .28 Asthenia Full sample 18.3 13.0 .28 Lithium 23.7 17.1 .48 Valproate 15.7 11.0 .42 Depression Full sample 17.9 17.4 Ͼ.99 Lithium 18.4 14.6 .8 Valproate 17.6 19.2 .85 Headache Full sample 15.7 18.3 .54 Lithium 13.2 22.0 .29 Valproate 17.0 16.4 Ͼ.99 Dizziness Full sample 13.5 7.0 .07 Lithium 7.9 9.8 .74 Valproate 16.3 4.1 .009 Diarrhea Full sample 11.8 14.8 .49 Lithium 11.8 19.5 .28 Valproate 11.8 12.3 Ͼ.99 Nervousness Full sample 10.5 14.8 .29 Lithium 10.5 19.5 .26 Valproate 10.5 12.3 .66 Thirst Full sample 10.0 6.1 .31 Lithium 5.3 7.3 .7 Valproate 12.4 5.5 .16 Speech disorder Full sample 6.6 0.9 .02 Lithium 7.9 0.0 .09 Valproate 5.9 1.4 .17

*Incidence in the olanzapine group greater than or equal to 10% or statistically significantly greater than placebo. †Treatment difference, olanzapine cotherapy vs monotherapy; derived from the Fisher exact test.

RISKS group analysis showed that, among patients receiving valproate, a significantly higher incidence of the adverse event No statistically significant changes from baseline were “dizziness” was seen in patients receiving cotherapy seen in extrapyramidal symptoms on the Simpson– (16.3% vs 4.1%; P=.009). Angus Scale, Abnormal Involuntary Movement Scale, and No statistically or clinically significant differences Barnes Akathisia Scale. Rates of adverse events (Table 4) emerged between the monotherapy and olanzapine co- more frequently reported in the cotherapy group in- therapy groups for changes in vital signs. However, the co- cluded somnolence, dry mouth, weight gain, increased therapy group experienced a 3.6% increase in body weight, appetite, tremor, and speech disorder (cotherapy: slurred significantly higher than that seen in the monotherapy group speech n=14, speaking difficulties n=1; monotherapy: (cotherapy: 3.08 [3.04] kg, n=219; monotherapy: 0.23 Ͻ stuttering n=1). In the cotherapy group, 25 patients dis- [2.48] kg, n=113; F1,302=73.88; P .001). With the excep- continued treatment due to adverse events (Table 1). Six tion of a greater incidence of treatment-emergent elevated (2.6%) discontinued due to somnolence, 3 (1.3%) due prolactin levels (upper limit: 0.81 nmol/L for men, 1.05 to weight gain, and 3 (1.3%) due to peripheral edema. nmol/L for women) at end point in the cotherapy group The remaining 13 discontinuing patients in the olanza- (19.1% vs 4.3%; P=.001), there were no other statisti- pine cotherapy group withdrew due to 13 different ad- cally and clinically significant differences in treatment- verse events (1 patient per event class). The 2 patients emergent laboratory test result abnormalities at end point, (1.7%) in the monotherapy group who discontinued both including nonfasting glucose levels, between the olanza- withdrew due to depression (Table 1). Post hoc sub- pine cotherapy group and the monotherapy group.

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 59, JAN 2002 WWW.ARCHGENPSYCHIATRY.COM 67

©2002 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/30/2021 COMMENT include the presence of depressive symptoms may particularly benefit from the addition of olanzapine. Overall YMRS responses to cotherapy with valproate This double-blind, placebo-controlled study suggests that, were similar in magnitude to responses to cotherapy with in patients with inadequate responses to at least 2 weeks lithium. On a numeric basis, response to lithium mono- of lithium or valproate monotherapy, the addition of olan- therapy was larger than the response to valproate mono- zapine may confer additional significant efficacy. No sig- therapy. This, plus the larger number of patients receiv- nificant changes were seen in extrapyramidal symp- ing valproate, might explain the significant difference toms but a number of adverse events were reported more between responses to valproate cotherapy vs valproate frequently by patients receiving cotherapy. monotherapy and the nonsignificant difference between re- The findings in this patient population, character- sponses to lithium cotherapy vs lithium monotherapy. ized as partially nonresponsive to monotherapy, are com- Adverse events occurring at a significantly higher rate parable with those of Mu¨ ller-Oerlinghausen et al,6 who in the cotherapy group and severe enough to result in dis- compared the efficacy of combined therapy with con- continuation included somnolence and weight gain. Im- ventional antipsychotics and valproate vs valproate mono- portantly, however, the olanzapine cotherapy group did therapy in patients with bipolar or schizoaffective dis- not experience a significant increase in nonfasting plasma order. Sachs18 recently reported preliminary results glucose levels or any treatment-emergent hyperglycemia. comparing the addition of risperidone, haloperidol, or Weight gain during treatment with olanzapine has been placebo to lithium or valproate, and found that the ad- described previously.20,21 However, lithium and valpro- dition of risperidone showed a significantly improved re- ate are both known to be associated with weight gain.22,23 sponse compared with monotherapy. Weight gain in the cotherapy group was similar to that re- Considering the exclusion of patients who showed a ported for olanzapine monotherapy, suggesting that there response after 2 weeks of monotherapy, our patient popu- is no clear synergism between olanzapine and either lithium lation should be classified as lithium- or valproate- or valproate in causing weight gain. nonresponders or partial responders. A significantly larger Our study has several limitations. First, assignment percentage of patients who received cotherapy with olan- to valproate or lithium was not randomized but reflected zapine achieved euthymia than did patients receiving val- the treatment preferences of clinicians and investigators. proate or lithium monotherapy. A potential limitation of The larger recruitment of patients receiving valproate this finding is the reliance on a single definition of eu- monotherapy reflects the current practice in the United thymia, which in this study was based on a YMRS total score States of a more extensive use of valproate than lithium less than or equal to 12 as suggested in the original pub- by patients who have bipolar disorder with manic or mixed lication of the scale.12 It is possible that another definition episodes. Because the study was not powered to show sig- may have led to substantially different remission rates. nificant differences in the primary outcome variables strati- One of the most interesting findings was the improve- fied by mood stabilizer, significant differences were found ment in depressive symptoms in this population of pa- only when comparisons were made between mono- tients with bipolar manic and mixed episodes. This was par- therapy and cotherapy; few differences were found when ticularly striking among patients with moderate-to-severe patients were stratified by mood stabilizer. Second, our symptoms of depression, who are particularly resistant to sample was restricted to partial responders to lithium or antimanic treatment.19 Among these patients, a nearly 7-fold valproate. Patients received monotherapy for only 2 weeks, greater improvement on the HAMD-21 scale was seen in with mean blood levels of 0.76 mmol/L and 63.6 µg/mL the cotherapy group. Furthermore, nearly 5 times as many for lithium and valproate, respectively. In clinical prac- patients in the cotherapy group showed at least a 50% im- tice, practitioners may be inclined to maximize the dose provement of depressive symptoms. These findings sug- of monotherapy treatment before initiating cotherapy treat- gest that cotherapy may provide substantial efficacy against ment. Finally, the lack of an olanzapine-monotherapy arm depressive symptoms in this patient population. prevented us from considering a possible synergistic effect Subgroup analysis also produced noteworthy find- between olanzapine and the mood stabilizer. Moreover, ings. In patients without psychotic features, cotherapy the lack of a comparator arm, such as a mood stabilizer was significantly superior to monotherapy. However, in plus another agent, also limits our conclusions about olan- those with psychotic features, the difference did not reach zapine’s unique effects when added to mood stabilizers. statistical significance, doubtlessly due to the greater sta- It would be of considerable benefit to investigate the ef- tistical power in the group of patients without psy- ficacy of olanzapine cotherapy vs a combination therapy chotic features. consisting of lithium plus valproate, thereby more accu- Patients with mixed episodes who received co- rately reflecting current clinical practice. therapy had a significantly larger response rate than did In summary, our findings suggest that, in patients those who received monotherapy. By contrast, patients with bipolar manic or mixed episodes who demonstrate who had purely manic episodes showed no significant inadequate responsiveness to at least 2 weeks of mood- difference between treatments. The response to co- stabilizer monotherapy, the combination of lithium or therapy was similar in magnitude in both patient types. valproate plus olanzapine may provide additional effi- However, patients with mixed episodes showed weaker cacy compared with either agent alone. Patients treated responses to monotherapy than did patients who were with combination therapy experienced more adverse purely manic. These results again suggest that patients events but none seemed to be life-threatening. The re- who had bipolar disorder whose illness characteristics sponse in patients without psychotic features and the im-

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 59, JAN 2002 WWW.ARCHGENPSYCHIATRY.COM 68

©2002 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/30/2021 Study Sites Principal Investigators versity of Texas Southwestern Medical Center, Dallas (Dr Suppes); Mood and Anxiety Disorders Program, Na- tional Institute of Mental Health, Bethesda, Md (Dr Zarate); North Bay Psychiatric Hospital, North Bay, Ontario: Department of Psychiatry, Case Western Reserve, Cleve- S. Adams, MD; The International Mood Center, La Jolla, land, Ohio (Dr Calabrese); Department of Psychiatry, Mas- Calif: H. S. Akiskal, MD; University of Pennsylvania, Phila- sachusetts General Hospital, Harvard Medical School, Bos- delphia, Pa: J. D. Amsterdam, MD; Princeton Biomedical Research, Princeton, NJ: J. T. Apter, MD; University of Texas ton (Dr Sachs); Department of Psychiatry, University of Health Science Center, San Antonio: C. L. Bowden, Jr, MD; Texas Health Science Center, San Antonio (Dr Bowden). Case Western Reserve University School of Medicine, Cleve- Corresponding author and reprints: Mauricio Tohen, land, Ohio: J. R. Calabrese, MD; McLean Hospital/ MD, DrPH, Lilly Research Laboratories, Eli Lilly & Co, Harvard Medical School, Belmont, Mass: F. Centorrino, Indianapolis, IN 46285. MD; Western Psychiatric Institute and Clinic/University of Pittsburgh Medical Center, Pittsburgh, Pa: K. N. R. Chen- gappa, MD; Reno, Nev: M. De Priest, MD; Falls Church, REFERENCES Va: D. G. Daniel, MD; Clinical Investigation Specialists Inc, Little Rock, Ark: B. Diner, MD; Pharmacology Re- 1. Sachs GS, Printz DJ, Kahn DA, Carpenter D, Docherty JP. The expert consensus search Corporation, Salt Lake City, Utah: J. M. Fergu- guideline series: medication treatment of bipolar disorder. Postgrad Med. 2000; Spec No:1-104. son, MD; Polyclinique Saint Laurent, Quebec City, PQ: 2. Swann A. Practical management of depressive and manic episodes. In: Garza- M. J. Filteau; Riverview Consultants, Calgary, Alberta: C. Trevino ES, ed. Medical Psychiatry: Theory and Practice. Vol 1. Teaneck, NJ: World Gorman, MD; Baltimore Veterans Affairs Medical Cen- Scientific; 1989. ter, Baltimore, Md: P. Hauser, MD; St Luke’s Roosevelt Out- 3. Freeman MP, Stoll AL. Mood stabilizer combinations: a review of safety and efficacy. Am J Psychiatry. 1998;155:12-21. patient Clinic, New York, NY: S. E. Hyler, MD; Monte- 4. Kane JM. The role of neuroleptics in manic-depressive illness. J Clin Psychiatry. fiore Medical Center/Albert Einstein College of Medicine, 1988;49(suppl):12-14. New York: N. Iqbal, MD; Center for Addiction and Mental 5. Klein E, Bental E, Lerer B, Belmaker RH. Carbamazepine and haloperidol vs pla- Health, Toronto, Ontario: S. Kennedy, MD; Stanford Uni- cebo and haloperidol in excited psychoses: a controlled study [abstract]. Am J Psychiatry. 1984;137:782-790. versity School of Medicine, Stanford, Calif: T. A. Ketter, 6. Mu¨ller-Oerlinghausen B, Retzow A, Henn FA, Giedke H, Walden J. Valproate as MD; Clinic Bois de Boulogne, Montreal, Quebec: J. Leblanc, an adjunct to neuroleptic medication for the treatment of acute episodes of ma- MD; Birmingham Psychiatric Pharmaceutical Studies, nia: a prospective, randomized, double-blind, placebo-controlled, multicenter study. Birmingham, Ala: H. E. Logue, MD; University of Cincin- J Clin Psychopharmacol. 2000;20:195-203. 7. Tohen M, Sanger TM, McElroy SL, Tollefson GD, Chengappa KNR, Daniel DG, nati Medical Center, Cincinnati, Ohio: S. L. McElroy, MD; Petty F, Centorrino F, Wang R, Grundy SL, Greaney MG, Jacobs TG, David SR, University of New Mexico Health Sciences Center, Toma V. Olanzapine vs placebo in the treatment of acute mania. Am J Psychia- Albuquerque: H. G. Nurnberg; Veterans Affairs Medical Cen- try. 1999;156:702-709. ter, Dallas, Tex: F. Petty, MD, PhD; Mesa Vista Hospital, 8. Tohen M, Jacobs TG, Grundy SL, Banov MC, McElroy SL, Janicak PG, Zhang F, Toma V, Francis BJ, Sanger TM, Tollefson GD, Breier A. Efficacy of olanzapine in San Diego, Calif: M. Plopper, MD; Bio Behavioral Research acute bipolar mania: a double-blind, placebo-controlled study. Arch Gen Psy- Center, Decatur, Ga: R. A. Riesenberg, MD; Massachusetts chiatry. 2000;57:841-849. General Hospital, Boston, Mass: G. S. Sachs, MD; Univer- 9. Zarate C Jr, Narendran R, Tohen M, Greaney JJ, Berman A, Pike S, Madrid A. sity of Texas Southwestern Medical Center, Dallas: T. Suppes, Clinical predictors of acute response with olanzapine in psychotic mood disorders. J Clin Psychiatry. 1998;59:24-28. MD, PhD; Crozer Chester Medical Center, Philadelphia: 10. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Dis- S. D. Targum, MD; University of Colorado Health Sciences orders, Fourth Edition. Washington, DC: American Psychiatric Association; 1994. Center, Denver, Colo: M. Thomas, MD; Medstream Inc, 11. First MB, Spitzer RL, Gibbon M, Williams JB. Structured Clinical Interview for Milwaukee, Wis: R. I. H. Wang, MD, PhD; Raleigh, NC: the Diagnostic and Statistical Manual, Fourth Edition, Patient Version. Washing- ton, DC: American Psychiatric Press; 1997. R. H. Weisler, MD; Psychiatric Institute of Florida, Orlando: 12. Young RC, Biggs JT, Ziegler VE, Meyer DA. A rating scale for mania: reliability, S. A. West, MD. validity, and sensitivity. Br J Psychiatry. 1978;133:429-435. 13. Hamilton M. A rating scale for depression. J Neurol Neurosurg Psychiatry. 1960; 23:56-62. 14. Guy W. ECDEU Assessment Manual for Psychopharmacology, Revised. Bethesda, Md: US Dept of Health, Education, and Welfare; 1976. provement of depressive symptoms suggests that the com- 15. Simpson GM, Angus JWS. A rating scale for extrapyramidal side effects. Acta bination of olanzapine and lithium or valproate may have Psychiatr Scand. 1970;212(suppl):S11-S19. mood-stabilizing properties in the acute treatment of bi- 16. Barnes TR. A rating scale for drug-induced akathisia. Br J Psychiatry. 1989;154: 672-676. polar manic or mixed episodes. 17. Gillings D, Koch G. The application of the principle of intention-to-treat to the analysis of clinical trials. Drug Inf J. 1991;25:411-425. 18. Sachs GS. Safety and efficacy of risperidone vs placebo as add-on therapy to Accepted for publication June 26, 2001. mood stabilizers in the treatment of the manic phase of bipolar disorder. Pre- This study was sponsored by Eli Lilly & Co, India- sented at the 2000 annual meeting of the American Psychiatric Association; May napolis, Ind. 17, 2000; Chicago, Ill. 19. Dilsaver SC, Swann AC, Shoaib AM, Bowers TC, Halle MT. Depressive mania as- This study was presented at the annual meeting of the sociated with nonresponse to antimanic agents. Am J Psychiatry. 1993;150: American Psychiatric Association, Chicago, Ill, May 15, 1548-1551. 2000. 20. Kelly DL, Conley RR, Love RC, Horn DS, Ushchak CM. Weight gain in adoles- cents treated with risperidone and conventional antipsychotics over six months. From the Department of Psychiatry, Harvard Medi- J Child Adolesc Psychopharmacol. 1998;8:151-159. cal School, McLean Hospital, Belmont, Mass (Dr Tohen); 21. Tollefson GD, Beasley CM Jr, Tran PV, Street JS, Krueger JA, Tamura RN, Graffeo KA, Thieme ME. Olanzapine vs haloperidol in the treatment of schizophrenia and Lilly Research Laboratories, Indianapolis, Ind (Drs Tohen, schizoaffective and schizophreniform disorders: results of an international col- Baker, Feldman, Tollefson, and Breier, and Mr Risser and laborative trial. Am J Psychiatry. 1997;154:457-465. Ms Keeter); Western Psychiatric Institute & Clinic, Uni- 22. Chen Y, Silverstone T. Lithium and weight gain. Int Clin Psychopharmacol. 1990; 5:217-225. versity of Pittsburgh Medical Center, Pittsburgh, Pa (Drs 23. Sachs GS, Guille C. Weight gain associated with use of psychotropic medica- Chengappa and Kupfer); Department of Psychiatry, Uni- tions. J Clin Psychiatry. 1999;60(suppl 21):16-19.

(REPRINTED) ARCH GEN PSYCHIATRY/ VOL 59, JAN 2002 WWW.ARCHGENPSYCHIATRY.COM 69